Artificial intelligence advertisement design and marketing recommendation platform based on big data

Abstract

The invention relates to the technical field of big data and artificial intelligence, and discloses an artificial intelligence advertisement design and marketing recommendation platform based on big data, which comprises a multi-source data acquisition module, a dynamic portrait modeling module, an AI advertisement generation engine, a cross-platform recommendation decision module and an effect tracking and compliance auditing module. By integrating multi-source data and utilizing an artificial intelligence algorithm to construct a label system, interests and demands of users can be accurately grasped, precise advertisement putting is realized, and the advertisement putting effect is improved; a personalized recommendation system generated through an AI advertisement generation engine can significantly improve the user satisfaction and participation degree, and through a cross-platform recommendation decision module, the distribution of advertisement resources is further optimized, and the overall marketing efficiency is improved; and the effect tracking and compliance auditing module ensures the transparency and legality of the advertisement activity, so that the advertisement putting is more accurate and effective.

Description

Technical Field

[0001] This invention relates to the fields of big data and artificial intelligence technology, specifically to an AI-based advertising design and marketing recommendation platform that integrates multi-source data and utilizes AI algorithms to achieve precise advertising design and efficient marketing recommendations. Background Technology

[0002] With the popularization of the internet and the rapid development of network application technologies, users' browsing history online can be analyzed and tracked. The use of this data by businesses or third-party service providers for marketing services such as consulting, strategy, and ad placement is known as big data marketing. Big data is likened to a vital natural resource, comparable to steam, electricity, and oil. It changes the way people think, make decisions, and act, making society more intelligent and optimizing corporate marketing decisions, thus being seen by businesses as the foundation of future competitive advantage.

[0003] However, existing advertising and marketing methods have many problems. For example, traditional platforms have difficulty integrating user behavior data from multiple channels such as e-commerce, short videos, and social media, resulting in one-sided user profiles. On the other hand, advertising creatives have a low degree of matching with users' real-time interests and lack the ability to dynamically allocate budgets across platforms. Summary of the Invention

[0004] The purpose of this invention is to provide an AI-powered advertising design and marketing recommendation platform based on big data, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an AI-based advertising design and marketing recommendation platform based on big data, comprising a multi-source data acquisition module, a dynamic profile modeling module, an AI advertising generation engine, a cross-platform recommendation decision module, and an effect tracking and compliance audit module; the multi-source data acquisition module is used to collect multi-source user data; the dynamic profile modeling module constructs a comprehensive model of user behavior and preferences based on the collected data; the AI ​​advertising generation engine automatically designs and generates personalized advertising content based on this model; the cross-platform recommendation decision module intelligently matches user profiles with advertising content to ensure that advertisements are placed on the most suitable platforms and time periods; the effect tracking and compliance audit module is responsible for monitoring the real-time effect of advertising placement, performing data analysis to optimize advertising strategies, and simultaneously reviewing the legality of advertising content to ensure compliance during the advertising placement process, thereby providing users with a more accurate and personalized marketing experience.

[0006] Preferably, the multi-source data acquisition module collects multi-source heterogeneous data through an API interface, including but not limited to web browsing data, social media interaction data, geographic location information, consumption history, and other available relevant user information. After data acquisition, feature extraction, data cleaning, analysis, and mining are further performed through data preprocessing techniques. The data preprocessing includes a data cleaning unit, a data integration unit, a data transformation unit, a data reduction unit, a data dimensionality reduction unit, and a data aggregation unit to ensure the accuracy and usability of the data.

[0007] Preferably, the multi-source data acquisition module supports dual-channel access of real-time streaming data (Kafka) and offline data (HDFS), supports storage and processing of data volumes up to PB level, can cope with high-concurrency requests in the context of big data, and ensures stable operation and efficient response of the platform.

[0008] Preferably, the dynamic profile modeling module constructs a tag system based on basic attributes, interests, and spending power, using an inverted index for storage. When constructing the tag system, a clustering algorithm is used to segment users. The clustering results categorize different user groups into clusters with similar characteristics, ensuring high similarity among users within each cluster. The formula is as follows: , in, It is the number of clusters. It is the first Clusters, It is the first The centroid (mean vector) of each cluster. User feature vector With cluster centroid Distance metric between them.

[0009] Preferably, when storing the inverted index, a mapping relationship is established between user IDs and tags, an inverted index table is constructed, and a list of user IDs corresponding to each tag is recorded.

[0010] Preferably, the dynamic portrait modeling module can also introduce a time decay factor, using the following formula: ,in: w(t) is the behavior weight at time t; λ is the decay rate, controlling how quickly the weight decreases over time; t is the time difference between the time the behavior occurred and the current time; real-time tracking and weight adjustment of user behavior enable the model to capture recent changes in user behavior. When calculating user behavior weights, the time decay factor can be applied to the original behavior weights. ,in: These are the adjusted behavior weights. This refers to the original behavior weights; after introducing a time decay factor, this module can also utilize a time window sliding mechanism. Assuming the time window length is T, the time window can be represented as: ,in: t is the current time, T is the length of the time window, and user behavior data is dynamically updated. Assuming the user behavior data is D(t), the behavior data within a time window can be represented as: ,in: It is user behavior data within a time window. This refers to user behavior data at time t. After the model training is complete, the dynamic profile modeling module will generate personalized feature vectors for each user based on the behavior data within the time window. and adjusted behavior weights Generate the user's personalized feature vector v: ,in: Each feature It is calculated based on user behavior data and weights, specifically using statistical methods such as weighted average and summation. Through these personalized feature vectors, the system can provide users with customized services and recommendations, such as personalized content pushes and product recommendations. Based on the generated personalized feature vector v, the system provides users with customized services and recommendations, and the recommendation formula can be expressed as: ,in: It is a recommendation algorithm function that generates recommendation results based on the user's personalized feature vector v.

[0011] Preferably, the AI ​​advertising generation engine generates image / video advertising creatives based on the GPT-4 multimodal model. Considering the user's personalized feature vector v, the advertising creatives are highly consistent with the user's preferences. The matching degree of the advertising content is evaluated through a specific scoring mechanism. The higher the score, the better the fit between the advertising creative and the user's feature vector. Finally, the generated recommendation results will be displayed through the user interface, aiming to improve the user experience and drive user engagement.

[0012] Preferably, the cross-platform recommendation decision module uses reinforcement learning algorithms to allocate advertising budgets for channels such as Douyin, WeChat, and Taobao, and monitors click-through rate and return on investment in real time and provides feedback for optimization.

[0013] Preferably, the effect tracking and compliance audit module includes data tracking, effect evaluation, and regulatory compliance to ensure the transparency and legality of advertising campaigns. This module can collect advertising exposure and conversion data across the entire chain through data tracking, and at the same time ensure data privacy and compliance through blockchain notarization, so that each advertising campaign can accurately reach the target user group, while complying with relevant laws and regulations to ensure the compliance of advertising content.

[0014] This invention provides an AI-powered advertising design and marketing recommendation platform based on big data. It offers the following advantages: (1) By integrating multi-source data and using artificial intelligence algorithms to construct a tag system, this invention can accurately grasp users' interests and needs, achieve precise advertising placement, and improve advertising effectiveness. The personalized recommendation system generated by the AI ​​advertising generation engine can significantly improve user satisfaction and participation. The cross-platform recommendation decision module further optimizes the allocation of advertising resources and improves overall marketing efficiency. The effect tracking and compliance audit module ensures the transparency and legality of advertising activities, making advertising placement more accurate and effective.

[0015] (2) By introducing a time decay factor, the present invention tracks and adjusts the weights of user behavior in real time, enabling the model to capture the recent changes in user behavior. After introducing the time decay factor, the module can also dynamically update user behavior data through a time window sliding mechanism. After the model training is completed, the dynamic profile modeling module will generate personalized feature vectors for each user. Through these personalized feature vectors, the system can provide users with customized services and recommendations, realizing a deeper level of personalized experience. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the process of an AI-based advertising design and marketing recommendation platform based on big data, according to the present invention. Detailed Implementation

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0018] Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the invention, and should not be construed as limiting the invention.

[0019] A preferred embodiment of the big data-based artificial intelligence advertising design and marketing recommendation platform provided by this invention is as follows: Figure 1 The diagram illustrates an AI-powered advertising design and marketing recommendation platform based on big data. It includes a multi-source data acquisition module, a dynamic profile modeling module, an AI advertising generation engine, a cross-platform recommendation decision-making module, and an effect tracking and compliance auditing module. The multi-source data acquisition module collects multi-source user data. The dynamic profile modeling module constructs a comprehensive model of user behavior and preferences based on the collected data. The AI ​​advertising generation engine automatically designs and generates personalized advertising content based on this model. The cross-platform recommendation decision-making module intelligently matches user profiles with advertising content to ensure that ads are placed on the most suitable platforms and time slots. The effect tracking and compliance auditing module monitors the real-time effects of advertising placement, performs data analysis to optimize advertising strategies, and conducts legality reviews of advertising content to ensure compliance during the advertising process, thereby providing users with a more accurate and personalized marketing experience. The multi-source data acquisition module collects heterogeneous data from multiple sources through API interfaces, including but not limited to web browsing data, social media interaction data, geographic location information, consumption history, and other available relevant user information. After data acquisition, feature extraction, data cleaning, analysis, and mining are further performed through data preprocessing techniques. Among them, data preprocessing includes data cleaning unit, data integration unit, data transformation unit, data reduction unit, data dimensionality reduction unit, and data aggregation unit to ensure the accuracy and usability of the data. The multi-source data acquisition module supports dual-channel access of real-time streaming data (Kafka) and offline data (HDFS), supports storage and processing of data volumes up to PB level, can cope with high-concurrency requests in the context of big data, and ensures stable operation and efficient response of the platform; The dynamic user profile modeling module constructs a tag system based on basic attributes, interests, and spending power, using an inverted index for storage. During tag system construction, a clustering algorithm is employed to segment users, grouping different user groups into clusters with similar characteristics to ensure high similarity within each cluster. The formula is as follows: , in, It is the number of clusters. It is the first Clusters, It is the first The centroid (mean vector) of each cluster. User feature vector With cluster centroid Distance metric between; When storing the inverted index, a mapping relationship is established between user IDs and tags, and an inverted index table is constructed to record the list of user IDs corresponding to each tag; The dynamic portrait modeling module can also introduce a time decay factor, using the following formula: ,in: w(t) is the behavior weight at time t; λ is the decay rate, controlling how quickly the weight decreases over time; t is the time difference between the time the behavior occurred and the current time; real-time tracking and weight adjustment of user behavior enable the model to capture recent changes in user behavior. When calculating user behavior weights, the time decay factor can be applied to the original behavior weights. ,in: These are the adjusted behavior weights. This refers to the original behavior weights; after introducing a time decay factor, this module can also utilize a time window sliding mechanism. Assuming the time window length is T, the time window can be represented as: ,in: t is the current time, T is the length of the time window, and user behavior data is dynamically updated. Assuming the user behavior data is D(t), the behavior data within a time window can be represented as: ,in: It is user behavior data within a time window. This refers to user behavior data at time t. After the model training is complete, the dynamic profile modeling module will generate personalized feature vectors for each user based on the behavior data within the time window. and adjusted behavior weights Generate the user's personalized feature vector v: ,in: Each feature It is calculated based on user behavior data and weights, specifically using statistical methods such as weighted average and summation. Through these personalized feature vectors, the system can provide users with customized services and recommendations, such as personalized content pushes and product recommendations. Based on the generated personalized feature vector v, the system provides users with customized services and recommendations, and the recommendation formula can be expressed as: ,in: It is a recommendation algorithm function that generates recommendation results based on the user's personalized feature vector v; The AI ​​ad generation engine generates image / video ad creatives based on the GPT-4 multimodal model. Taking into account the user's personalized feature vector v, the ad creatives are highly consistent with the user's preferences. The matching degree of the ad content is evaluated through a specific scoring mechanism. The higher the score, the better the fit between the ad creative and the user's feature vector. Finally, the generated recommendation results will be displayed through the user interface, aiming to improve the user experience and drive user engagement. The cross-platform recommendation decision module uses reinforcement learning algorithms to allocate advertising budgets for channels such as Douyin, WeChat, and Taobao, and monitors click-through rate and return on investment in real time and provides feedback for optimization. The performance tracking and compliance audit module includes data tracking, performance evaluation, and regulatory compliance, ensuring the transparency and legality of advertising campaigns. This module can collect data on ad exposure and conversion across the entire process through event tracking, while using blockchain for data storage to ensure data privacy and compliance. This allows each advertising campaign to accurately reach the target user group while adhering to relevant laws and regulations to ensure the compliance of advertising content.

[0020] In practice, taking an e-commerce platform as an example, data collection code is embedded in its website and mobile application to collect users' web browsing data, social media interaction data, geolocation information, purchase history, and other relevant user information. Simultaneously, user interaction data from social media platforms, such as likes, comments, and shares, is integrated to form a comprehensive user behavior dataset. The collected data undergoes feature extraction, data cleaning, analysis, and mining using data preprocessing techniques. Data from different data sources is integrated and transformed, and data formats and encoding rules are standardized. For example, basic information such as users' age and gender is standardized, and users' purchasing behavior data is discretized for subsequent analysis and modeling.

[0021] User features are extracted from dimensions such as basic attributes, interests and preferences, and spending power to construct feature vectors. For example: basic attributes: age, gender, region; interests and preferences: browsing history, click behavior, favorite content; spending power: purchase amount, purchase frequency, average order value; feature values ​​are scaled to the range of [0, 1] to avoid the influence of different units of measurement; PCA (principal component analysis) or t-SNE is used to reduce feature dimensions and improve clustering efficiency; users are segmented using clustering algorithms (such as K-Means) to determine the number of clusters k, and labels are generated for each cluster to describe the common characteristics of users within the cluster. For example: cluster 1: high spending power, preference for technology products, users in first-tier cities; cluster 2: low spending power, preference for entertainment content, users in second-tier cities. An inverted index is used for storage. When storing the inverted index, a mapping relationship is established between user IDs and tags, and an inverted index table is constructed to record the list of user IDs corresponding to each tag. Image / video ad creatives are generated based on the GPT-4 multimodal model. Considering the user's personalized feature vector v, the ad creatives are highly consistent with user preferences. The matching degree of ad content is evaluated through a specific scoring mechanism. The higher the score, the better the fit between the ad creative and the user's feature vector. Finally, the generated recommendation results will be displayed through the user interface, aiming to improve user experience and drive user engagement. Reinforcement learning algorithms are used to allocate the advertising budget for channels such as Douyin, WeChat, and Taobao, and click-through rate and ROI are monitored in real time and feedback is provided for optimization. Data on ad exposure and conversion throughout the entire process is collected through event tracking, and data privacy and compliance are ensured through blockchain notarization. This ensures that each advertising campaign can accurately reach the target user group while complying with relevant laws and regulations to ensure the compliance of ad content.

[0022] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0023] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A big data-based artificial intelligence advertising design and marketing recommendation platform, characterized in that, It includes a multi-source data acquisition module, a dynamic profile modeling module, an AI ad generation engine, a cross-platform recommendation decision-making module, and an effect tracking and compliance auditing module; the multi-source data acquisition module is used to collect multi-source data of users; the dynamic profile modeling module constructs a comprehensive model of user behavior and preferences based on the collected data; and the AI ​​ad generation engine automatically designs and generates personalized ad content based on this model. The cross-platform recommendation decision-making module intelligently matches user profiles with ad content to ensure that ads are delivered to the most suitable platforms and time periods. The performance tracking and compliance audit module is responsible for monitoring the real-time performance of ad delivery, conducting data analysis to optimize ad strategies, and reviewing the legality of ad content to ensure compliance during ad delivery, thereby providing users with a more accurate and personalized marketing experience.

2. The AI-powered advertising design and marketing recommendation platform based on big data as described in claim 1, characterized in that: The multi-source data acquisition module collects heterogeneous data from multiple sources through API interfaces, including but not limited to web browsing data, social media interaction data, geographic location information, consumption history, and other available relevant user information. After data acquisition, feature extraction, data cleaning, analysis, and mining are further performed through data preprocessing techniques. Among them, data preprocessing includes a data cleaning unit, a data integration unit, a data transformation unit, a data reduction unit, a data dimensionality reduction unit, and a data aggregation unit to ensure the accuracy and usability of the data.

3. The AI-powered advertising design and marketing recommendation platform based on big data as described in claim 1, characterized in that: The multi-source data acquisition module supports dual-channel access for real-time streaming data (Kafka) and offline data (HDFS), and supports storage and processing of data volumes up to PB level. It can cope with high-concurrency requests in the context of big data, ensuring the stable operation and efficient response of the platform.

4. The AI-powered advertising design and marketing recommendation platform based on big data as described in claim 1, characterized in that: The dynamic profile modeling module constructs a tag system based on basic attributes, interests, and spending power, using an inverted index for storage. During the tag system construction, a clustering algorithm is employed to segment users. The clustering results categorize different user groups into clusters with similar characteristics, ensuring high similarity among users within each cluster. The formula is as follows: ， in, It is the number of clusters. It is the first Clusters, It is the first The centroid (mean vector) of each cluster. User feature vector With cluster centroid Distance metric between them.

5. The AI-powered advertising design and marketing recommendation platform based on big data as described in claim 1, characterized in that: When storing the inverted index, a mapping relationship is established between user IDs and tags, and an inverted index table is constructed to record the list of user IDs corresponding to each tag.

6. The AI-powered advertising design and marketing recommendation platform based on big data as described in claim 1, characterized in that: The dynamic portrait modeling module can also introduce a time decay factor, using the following formula: ,in: w(t) is the behavior weight at time t; λ is the decay rate, controlling how quickly the weight decreases over time; t is the time difference between the time the behavior occurred and the current time; real-time tracking and weight adjustment of user behavior enable the model to capture recent changes in user behavior. When calculating user behavior weights, the time decay factor can be applied to the original behavior weights. ,in: These are the adjusted behavior weights. This refers to the original behavior weights; after introducing a time decay factor, this module can also utilize a time window sliding mechanism. Assuming the time window length is T, the time window can be represented as: ,in: t is the current time, T is the length of the time window, and user behavior data is dynamically updated. Assuming the user behavior data is D(t), the behavior data within a time window can be represented as: ,in: It is user behavior data within a time window. This refers to user behavior data at time t. After the model training is complete, the dynamic profile modeling module will generate personalized feature vectors for each user based on the behavior data within the time window. and adjusted behavior weights Generate the user's personalized feature vector v: ,in: Each feature It is calculated based on user behavior data and weights, specifically using statistical methods such as weighted average and summation. Through these personalized feature vectors, the system can provide users with customized services and recommendations, such as personalized content pushes and product recommendations. Based on the generated personalized feature vector v, the system provides users with customized services and recommendations, and the recommendation formula can be expressed as: ,in: It is a recommendation algorithm function that generates recommendation results based on the user's personalized feature vector v.

7. The AI-powered advertising design and marketing recommendation platform based on big data as described in claim 1, characterized in that: The AI ​​ad generation engine generates image / video ad creatives based on the GPT-4 multimodal model. Considering the user's personalized feature vector v, the ad creatives are highly consistent with the user's preferences. The matching degree of the ad content is evaluated through a specific scoring mechanism. The higher the score, the better the fit between the ad creative and the user's feature vector. Finally, the generated recommendation results will be displayed through the user interface, aiming to improve the user experience and drive user engagement.

8. The AI-powered advertising design and marketing recommendation platform based on big data according to claim 1, characterized in that: The cross-platform recommendation decision module uses reinforcement learning algorithms to allocate advertising budgets for channels such as Douyin, WeChat, and Taobao, and monitors click-through rate and return on investment in real time and provides feedback for optimization.

9. The AI-powered advertising design and marketing recommendation platform based on big data as described in claim 1, characterized in that: The performance tracking and compliance audit module includes data tracking, performance evaluation, and regulatory compliance, ensuring the transparency and legality of advertising campaigns. This module can collect advertising exposure and conversion data across the entire process through event tracking, while ensuring data privacy and compliance through blockchain notarization. This allows each advertising campaign to accurately reach the target user group, while adhering to relevant laws and regulations to ensure the compliance of advertising content.

Images